数据清洗及特征处理_随笔

数据清洗及特征处理数据清洗及特征处理

开始之前，导入numpy、pandas包和数据

#加载所需的库
import numpy as np
import pandas as pd

#加载数据train.csv
df = pd.read_csv('train.csv')
df.head(3)

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarked0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS1211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C2313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS 数据清洗简述

我们拿到的数据通常是不干净的，所谓的不干净，就是数据中有缺失值，有一些异常点等，需要经过一定的处理才能继续做后面的分析或建模，所以拿到数据的第一步是进行数据清洗，本章我们将学习缺失值、重复值、字符串和数据转换等 *** 作，将数据清洗成可以分析或建模的样子。

缺失值观察与处理

我们拿到的数据经常会有很多缺失值，比如我们可以看到Cabin列存在NaN，那其他列还有没有缺失值，这些缺失值要怎么处理呢

任务一：缺失值观察

(1) 请查看每个特征缺失值个数
(2) 请查看Age， Cabin， Embarked列的数据以上方式都有多种方式，所以建议大家学习的时候多多益善

#方法一
df.info()


RangeIndex: 891 entries, 0 to 890
Data columns (total 12 columns):
 #   Column       Non-Null Count  Dtype  
---  ------       --------------  -----  
 0   PassengerId  891 non-null    int64  
 1   Survived     891 non-null    int64  
 2   Pclass       891 non-null    int64  
 3   Name         891 non-null    object 
 4   Sex          891 non-null    object 
 5   Age          714 non-null    float64
 6   SibSp        891 non-null    int64  
 7   Parch        891 non-null    int64  
 8   Ticket       891 non-null    object 
 9   Fare         891 non-null    float64
 10  Cabin        204 non-null    object 
 11  Embarked     889 non-null    object 
dtypes: float64(2), int64(5), object(5)
memory usage: 83.7+ KB

#方法二
df.isnull().sum()

PassengerId      0
Survived         0
Pclass           0
Name             0
Sex              0
Age            177
SibSp            0
Parch            0
Ticket           0
Fare             0
Cabin          687
Embarked         2
dtype: int64

df[['Age','Cabin','Embarked']].head(3)

AgeCabinEmbarked022.0NaNS138.0C85C226.0NaNS

对缺失值进行处理

(1)处理缺失值一般有几种思路

(2) 请尝试对Age列的数据的缺失值进行处理

(3) 请尝试使用不同的方法直接对整张表的缺失值进行处理

以下是举例：

df[df['Age']==None]=0
df.head(3)

df[df['Age'].isnull()] = 0 # 还好
df.head(3)

df[df['Age'] == np.nan] = 0
df.head()

【思考】检索空缺值用np.nan,None以及.isnull()哪个更好，这是为什么？如果其中某个方式无法找到缺失值，原因又是为什么？

【回答】数值列读取数据后，空缺值的数据类型为float64所以用None一般索引不到，比较的时候最好用np.nan

df.dropna().head(3)

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarked1211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C3411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S5000000.00000.000000

df.fillna(0).head(3)

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarked0103Braund, Mr. Owen Harrismale22.010A/5 211717.25000S1211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C2313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.92500S

【思考】dropna和fillna有哪些参数，分别如何使用呢?

【参考】https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Dataframe.dropna.html

【参考】https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Dataframe.fillna.html

重复值观察与处理

由于这样那样的原因，数据中会不会存在重复值呢，如果存在要怎样处理呢

任务一：请查看数据中的重复值

df[df.duplicated()]

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarked17000000.00000.00019000000.00000.00026000000.00000.00028000000.00000.00029000000.00000.000.......................................859000000.00000.000863000000.00000.000868000000.00000.000878000000.00000.000888000000.00000.000

176 rows × 12 columns

2.2.2 任务二：对重复值进行处理

(1)重复值有哪些处理方式呢？

(2)处理我们数据的重复值

方法多多益善

以下是对整个行有重复值的清理的方法举例：

df = df.drop_duplicates()
df.head()

2.2.3 任务三：将前面清洗的数据保存为csv格式

df.to_csv('test_clear.csv')

特征观察与处理

我们对特征进行一下观察，可以把特征大概分为两大类：
数值型特征：Survived ，Pclass， Age ，SibSp， Parch， Fare，其中Survived， Pclass为离散型数值特征，Age，SibSp， Parch， Fare为连续型数值特征
文本型特征：Name， Sex， Cabin，Embarked， Ticket，其中Sex， Cabin， Embarked， Ticket为类别型文本特征。

数值型特征一般可以直接用于模型的训练，但有时候为了模型的稳定性及鲁棒性会对连续变量进行离散化。文本型特征往往需要转换成数值型特征才能用于建模分析。

2.3.1 任务一：对年龄进行分箱（离散化）处理

(1) 分箱 *** 作是什么？

(2) 将连续变量Age平均分箱成5个年龄段，并分别用类别变量12345表示

(3) 将连续变量Age划分为(0,5] (5,15] (15,30] (30,50] (50,80]五个年龄段，并分别用类别变量12345表示

(4) 将连续变量Age按10% 30% 50% 70% 90%五个年龄段，并用分类变量12345表示

(5) 将上面的获得的数据分别进行保存，保存为csv格式

#将连续变量Age平均分箱成5个年龄段，并分别用类别变量12345表示
df['AgeBand'] = pd.cut(df['Age'], 5,labels = [1,2,3,4,5])
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBand0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS21211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C32313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS23411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S34503Allen, Mr. William Henrymale35.0003734508.0500NaNS3

df.to_csv('test_ave.csv')

#将连续变量Age划分为(0,5] (5,15] (15,30] (30,50] (50,80]五个年龄段，并分别用类别变量12345表示
df['AgeBand'] = pd.cut(df['Age'],[0,5,15,30,50,80],labels = [1,2,3,4,5])
df.head(3)

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBand0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS31211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C42313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS3

df.to_csv('test_cut.csv')

#将连续变量Age按10% 30% 50 70% 90%五个年龄段，并用分类变量12345表示
df['AgeBand'] = pd.qcut(df['Age'],[0,0.1,0.3,0.5,0.7,0.9],labels = [1,2,3,4,5])
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBand0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS21211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C52313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS33411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S44503Allen, Mr. William Henrymale35.0003734508.0500NaNS4

df.to_csv('test_pr.csv')

【参考】https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.cut.html

【参考】https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.qcut.html

任务二：对文本变量进行转换

(1) 查看文本变量名及种类
(2) 将文本变量Sex， Cabin ，Embarked用数值变量12345表示
(3) 将文本变量Sex， Cabin， Embarked用one-hot编码表示

方法多多益善

#查看类别文本变量名及种类

#方法一: value_counts
df['Sex'].value_counts()

male      453
female    261
0           1
Name: Sex, dtype: int64

df['Cabin'].value_counts()

G6             4
C23 C25 C27    4
B96 B98        4
F33            3
C22 C26        3
              ..
D37            1
C92            1
E58            1
E77            1
B4             1
Name: Cabin, Length: 135, dtype: int64

df['Embarked'].value_counts()

S    554
C    130
Q     28
0      1
Name: Embarked, dtype: int64

#方法二: unique
df['Sex'].unique()

array(['male', 'female', 0], dtype=object)

df['Sex'].nunique()

#将类别文本转换为12345

#方法一: replace
df['Sex_num'] = df['Sex'].replace(['male','female'],[1,2])
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBandSex_num0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS211211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C522313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS323411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S424503Allen, Mr. William Henrymale35.0003734508.0500NaNS41

#方法二: map
df['Sex_num'] = df['Sex'].map({'male': 1, 'female': 2})
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBandSex_num0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS21.01211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C52.02313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS32.03411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S42.04503Allen, Mr. William Henrymale35.0003734508.0500NaNS41.0

#方法三: 使用sklearn.preprocessing的LabelEncoder
from sklearn.preprocessing import LabelEncoder
for feat in ['Cabin', 'Ticket']:
    lbl = LabelEncoder()  
    label_dict = dict(zip(df[feat].unique(), range(df[feat].nunique())))
    df[feat + "_labelEncode"] = df[feat].map(label_dict)
    df[feat + "_labelEncode"] = lbl.fit_transform(df[feat].astype(str))

df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFareCabinEmbarkedAgeBandSex_numCabin_labelEncodeTicket_labelEncode0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500NaNS21.01354091211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833C85C52.0744722313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250NaNS32.01355333411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000C123S42.050414503Allen, Mr. William Henrymale35.0003734508.0500NaNS41.0135374

#将类别文本转换为one-hot编码

#方法一: oneHotEncoder
for feat in ["Age", "Embarked"]:
#     x = pd.get_dummies(df["Age"] // 6)
#     x = pd.get_dummies(pd.cut(df['Age'],5))
    x = pd.get_dummies(df[feat], prefix=feat)
    df = pd.concat([df, x], axis=1)
    #df[feat] = pd.get_dummies(df[feat], prefix=feat)
    
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFare...Age_66.0Age_70.0Age_70.5Age_71.0Age_74.0Age_80.0Embarked_0Embarked_CEmbarked_QEmbarked_S0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500...00000000011211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833...00000001002313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250...00000000013411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000...00000000014503Allen, Mr. William Henrymale35.0003734508.0500...0000000001

5 rows × 109 columns

从纯文本Name特征里提取出Titles的特征(所谓的Titles就是Mr,Miss,Mrs等)

df['Title'] = df.Name.str.extract('([A-Za-z]+).', expand=False)
df.head()

PassengerIdSurvivedPclassNameSexAgeSibSpParchTicketFare...Age_66.0Age_70.0Age_70.5Age_71.0Age_74.0Age_80.0Embarked_CEmbarked_QEmbarked_STitle0103Braund, Mr. Owen Harrismale22.010A/5 211717.2500...000000001Mr1211Cumings, Mrs. John Bradley (Florence Briggs Th...female38.010PC 1759971.2833...000000100Mrs2313Heikkinen, Miss. Lainafemale26.000STON/O2. 31012827.9250...000000001Miss3411Futrelle, Mrs. Jacques Heath (Lily May Peel)female35.01011380353.1000...000000001Mrs4503Allen, Mr. William Henrymale35.0003734508.0500...000000001Mr

5 rows × 108 columns

# 保存上面的为最终结论
df.to_csv('test_fin.csv')

欢迎分享，转载请注明来源：内存溢出

原文地址: http://outofmemory.cn/zaji/5678808.html

数据清洗及特征处理

发表评论

评论列表（0条）